Human liver folylpolyglutamate synthetase: Biochemical characterization and interactions with folates and folate antagonists

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 256, No. 2, August 1, pp. 585-596, 1987

Human Liver Folylpolyglutamate Synthetase: Biochemical Characterization and Interactions with Folates and Folate Antagonists’

LYNN CLARKE AND DAVID J. WAXMAN’

Department of Biological Chemistry and Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts 02115

Received December 2,1986, and in revised form April 6,1987

Folylpolyglutamate synthetase (FPGS) was isolated from human liver cytosol by O- 30% (w/v) ammonium sulfate fractionation and characterized biochemically. Using am- imopterin (AMT), L-[3Hlglutamate and MgATP as cosubstrates, maximal y-L-glutamylation activity was observed in the presence of the activators KC1 and NaHC03. ATP and 2-mercaptoethanol were each required for enzyme activity and stability. In the absence of ATP, human liver FPGS rapidly inactivated at 37°C (ti,z - 8 min), whereas FPGS isolated from rabbit liver was significantly more stable (tllz = 68 min). Both folates and antifolates were effectively polyglutamylated by the isolated human liver enzyme. Km parameters determined for AMT (K, = 4.3 PM) were similar to those determined for several reduced folates (tetrahydrofolic acid, dihydrofolic acid, and folinic acid; Km = 3-7 FM), while significantly higher Km values were observed for methotrexate (MTX) and 5-methyltetrahydrofolic acid (Km = 50-60 PM) and for folic acid (K, = 100 PM). All of the substrates examined exhibited V,,, values ranging from 30 to 90% of the AMT value (V,,, = 935 pmol product/mg/h). The order of reactivity for these substrates differed from that determined in parallel studies for FPGS isolated from rat and rabbit liver. In the case of AMT and several reduced folates, inhibition of human liver FPGS was observed at substrate concentrations at or above 50-250 PM. FPGS isolated from six individual human livers exhibited highly similar biochemical and kinetic properties, suggesting the presence of the same or at least highly similar enzyme species in each individual, with a five-fold interindividual range in specific activities observed. Comparison of MTX with its higher polyglutamates (MTX-Glu2 to MTX-Glu& as FPGS substrates indicated a significant decrease in V,,, values with increasing glutamate chain length which was partially compensated for by a corresponding decrease in K,. Consistent with these observations, the isolated enzyme was unable to synthesize polyglutamates higher than MTX-Glu3 when MTX was supplied as substrate, raising the question as to how MTX polyglutamates containing up to five or six y-L-glutamate residues are formed in vivo. 0 1987 Academic Press. Inc.

Tetrahydrofolic acid and related folate coenzymes are almost always isolated from natural sources as their oligo-y-L-glutamyl

r Supported in part by Grants CA19589 and RR05526 from the National Institutes of Health and a grant from the Research Corporation (D.J.W.). Human liver samples used in this study were obtained from the Nashville Organ Procurement Agency through NIH Grant ES00267 to Dr. F. Peter Guengerich, Vanderbilt 1~Jniversity.

derivatives (1). The polyglutamate chain length distribution is dependent on the species, the folate, and possibly also on the tissue (2, 3), with mammalian folylpolyglutamates typically containing two to six y-L-glutamyl residues (4). Such species

’ To whom correspondence should be addressed at Dana-Farber Cancer Institute, 44 Binney Street, Bos- ton, MA 02115.

585 0003-9861/87 $3.00 Copyright 0 198’7 by Academic Press, Inc. All rights of reproduction in any form reserved.

586 CLARKE AND WAXMAN

are utilized in the cell as substrates for all folate-dependent biosynthetic reactions. Polyglutamylation facilitates the retention of folates transported into the cell (5) and folylpolyglutamates can inhibit and thereby regulate folate-dependent reactions for which they are not substrates (1).

The cytosolic enzyme folylpolyglutamate synthetase (FPGS)3 catalyzes the ATP-dependent y-L-glutamylation of folates, ac- cepting a wide variety of substrates including folate antagonists such as aminopterin (AMT) and methotrexate (MTX). Polyglutamylation of MTX is an essential determinant of its cytotoxicity since polyglutamylation is necessary for intracellu- lar drug retention (6) and the polyglutamates of MTX are at least as inhibitory to dihydrofolate reductase as is MTX itself (7). Polyglutamylation of MTX also affects the selectivity of leucovorin (folinie acid) rescue (8) and enhances by up to lOOO-fold the effectiveness of MTX as an inhibitor of other folate-dependent enzymes such as thymidylate synthetase and 5-aminoimid- azole carboxamide ribotide tranformylase (9,lO).

FPGS has been purified to homogeneity from bacteria (11-13) and from hog liver (14) and has been partially purified from Chinese hamster ovary cells (15) and from rat (16), mouse (1’7), and beef (18) liver. Al- though these latter mammalian FPGS preparations show broadly similar biochemical properties, important interspe- ties differences are emerging, indicating the need for detailed biochemical studies of the corresponding human liver enzyme. Thus far, however, studies of polyglutamy-

3 Abbreviations used: FPGS, folylpolyglutamate synthetase; HLn, human liver sample prepared from individual 12 as described under Materials and Meth- ods; MTX, methotrexate (4-amino-4-deoxy-N”-meth- ylpteroyl-L-glutamic acid); AMT, aminopterin (4- amino-4-deoxypteroyl-L-glutamic acid); folic acid, pteroyl-L-glutamic acid; folinic acid, 5-formyltetra- hydrofolate; MTX-G,, methotrexate polyglutamate where n is the total number of y-L-glutamyl residues (thus MTX-G1 refers to methotrexate, MTX-G2 to MTX after incorporation of one additional y-L-glutamyl residue, etc.); Hepes, 4-(2-hydroxyethyl)-l-piperaz- ineethanesulfonic acid; 2-ME, 2-mercaptoethanol.

lation in humans have largely been con- fined to whole cells [e.g. (2,5,19,20)] where interpretation of the results is complicated by transport effects and by the subsequent metabolism of any polyglutamates formed. The current studies were therefore under- taken to address several questions, including: (1) Can FPGS be isolated from fresh human tissue sources, such as liver, in an active form? Are the polyglutamylation activities comparable to those observed in laboratory animals? In the case of dihydrofolate reductase, another enzyme of folate metabolism, activity levels measured in human liver samples are reported to be low to undetectable (21). (2) How similar is human liver FPGS to FPGS isolated from livers of other mammalian species? In particular, does the human liver enzyme exhibit the broad specificity for both folates and antifolates reported for the rat and mouse liver enzymes? Are findings obtained in studies of these latter enzymes generally applicable to the human enzyme? Finally (3) are there important differences in the properties of FPGS isolated from individual human subjects? These and related questions are addressed in the present study, which reports the isolation and biochemical characterization of FPGS from multiple human liver samples, its comparison to enzyme isolated from rat and rabbit liver, and its specificity for folates, antifolates, and antifolate oligoglutamates as substrates in the y+glutamylation reaction.

MATERIALS AND METHODS

Mutericcls. MTX was obtained from the National Cancer Institute and AMT was from Southern Re- search Institute (Birmingham, AL). Unlabeled polyglutamates of MTX (MTX-G2 to MTX-G6) were pur- chased from American Radiolabeled Chemicals (St. Louis, MO) and, with the exception of MTX-GI, were shown to be homogeneous when analyzed by HPLC as described below. The MTX-GI contaminant eluted from HPLC following MTX-G7 (obtained from Dr. M. G. Nair, University of South Alabama, Mobile, AL) but prior to MTX-GG and accounted for about 20% of the total Am of the MTX-GI sample. Folic acid and tetrahydrofolic acid were obtained from Dr. J. Wright of this institute. Folinic acid, dihydrofolic acid, 5- methyltetrahydrofolie acid, ATP, AMP, and DEAE- cellulose (coarse mesh) were from Sigma, and dATP,

HUMAN LIVER FOLYLPOLYGLUTAMATE SYNTHETASE 587

ADP, dGTP, UTP, and dTTP were from P-L Biochem- icals. Purity of the reduced folates was assessed by HPLC and was shown to be -90% (Cl8 chromatography with a linear gradient of 0 to 10% acetonitrile in 0.1 M ammonium acetate, pH 7.8, over ‘70 min).

Human livers from organ donors who met acciden- tal deaths were obtained from the Nashville Organ F’rocurement Agency. Individual human livers were designated HL33 (25 year, 8) HL35 ( 28 year, b), HL36 (32 year, b), HL37 (15 year, P), HL38 (42 year, b), and FIL39 (24 year, 6), with the ages and sexes of the donors as indicated. Livers (1.5 to 2 kg, wet wt) were removed and perfused with a cold saline solution within 30 min of clinical death and then frozen in small portions (1 to 6 cma) in liquid nitrogen for storage at -80°C. Samples were then thawed and homogenized in 0.1 M Tris acetate, pH 7.4, containing 0.1 M KCl, 1 mM eth- ylenediaminetetraacetate, 0.4 mM phenylmethylsul- fonyl fluoride, and 20 pM butylated hydroxytoluene. Cytosolic fractions were prepared by ultracentrifugation and then stored at -80°C. Human liver cytosolic fractions were kindly prepared and provided by Dr. F. Peter Guengerich, Vanderbilt University. Cy- tosolic fractions were prepared as described above from fresh, saline-perfused livers obtained from adult male Sprague-Dawley rats (Charles River Breeding Laboratories, Wilmington, MA) or from adult female New Zealand White rabbits (Pine Acres Rabbitry, W. Brattleboro, VT) by homogenization in 10 mM TrisCl, pH 7.8, containing 0.25 M sucrose and 1 mM ethylene- diaminetetraacetate followed by ultracentrifugation.

Puri~cation of ~[~Hlglutamic acid. L-[G-‘H]Glu- tamic acid (Amersham) with a specific activity of 39 Ci/mmol (1 mCi in 1 ml of 25 mM sodium acetate buffer, pH 5.2) was purified by application to a column of DEAE-cellulose (0.75 X 2.8 cm) which was eluted with 25 mM sodium acetate, pH 5.2 (40 ml), followed by elution with 0.1 M HCI (2.5 ml). Fractions eluted with the sodium acetate buffer and containing the majority of the applied cpm (-10 ml, 98% of applied activity) were combined and then mixed with unlabeled L-glutamate to give a final specific activity of 5 or 20 mCi/mmol. This purification procedure reduced the background radioactivity eluted with the 0.1 M

HCl wash from -0.5% to ~0.01% of the total applied cpm.

Enzyme preprutim Cytosol prepared from human, rat, or rabbit liver was stirred at 0°C and solid ammonium sulfate was added to 30% saturation (176 g added per liter of sample) over 30 min. The mixture was stirred on ice for a further 30 min and then the precipitate was collected by centrifugation at 10,000~ for 15 min. The pellet was suspended to about 10% of the original volume of cytosol with 20 mM Hepes, pH 7.4, containing 0.25 M sucrose and 50 mM 2-mercaptoethanol (2-ME) and then centrifuged at 40,OOOg for 45 min. The supernatant thus obtained was dialyzed for about 20 h against the same buffer (3 changes of

10 to 15 vol each). Precipitate which formed during the dialysis was removed by centrifugation and the resulting supernatant (6-12 mg protein/ml) was ali- quoted and stored at -20°C. Initial activities of human liver cytosols ranged from 65 to 280 pmol/mg/h (133 + 75, n = 6) with activities of 300 to 1300 pmol/mg/ h (738 -C 350, ?z = 6) obtained for the corresponding isolated enzyme fractions, giving a fold purification of 6.3 f 2.1, measured with AMT as FPGS substrate.

FPGS assays. Standard assay mixtures contained isolated enzyme (2 mg protein/ml), 1 mM L- [aH]glutamate (5 mCi/mmol) (added via 20 ~125 mM sodium acetate, pH 5.2), 500 pM AMT, MTX, or other (anti)folate, and 0.2M TrisCl (pH 8.3 at 37”C), containing 30 mM KCI, 25 mM MgClz, 30 mM NaHCO,, and 45 mM 2-ME in a final volume of 0.25 ml. Reactions were initiated by the addition of NazATP to 5 mM to prewarmed samples and the samples were then incubated for 1 h at 37°C. The reaction was stopped by dilution with 0.75 ml of 93 mM sodium acetate, pH 5.2, and then the tubes were placed on ice. The triti- ated product was separated from unreacted L- [3Hlglutamate using small DEAE-cellulose columns (0.75 X 1.4 cm) by a modification of the method of Ref. (18). (DEAE-cellulose was swelled then washed three times with 1 M sodium phosphate, pH 8.9, and then washed extensively with distilled water; the resin was discarded after each use.) Columns were washed with 40 ml 70 mM Na acetate, pH 5.2 (flow rate = 1 ml/ min) and then the radioactive product was eluted di- rectly into scintillation vials with 2.5 ml 0.1 M HCl. Hydrofluor (10 ml; National Diagnostics) was added and the mixture was shaken thoroughly then allowed to stand for at least 1 h before counting in a Beckman LS7000 liquid scintillation counter at a ‘H counting efficiency of -40% .Under standard assay conditions, 40 to 90 cpm (-0.005% of total applied cpm) were eluted from the DEAE-cellulose column in the 0.1 M NC1 wash when either enzyme, ATP, or both were absent from the incubation mixture, while complete incubation mixtures typically yielded 500 to 1500 cpm. For each assay series, representative samples without ATP were prepared in duplicate, and the mean value of cpm eluted for those samples was used as a background value when calculating product formation for the other samples. FPGS activity was linear with re- spect to protein from 0 to 3.2 mg protein/ml. Activities were also shown to be linear for at least 2 h when using AMT, MTX, and folinic acid as substrates at concentrations equal to either 5X or 0.4X their re- spective Km values.

Recovery of MTX-Ga in the DEAE-cellulose 0.1 M

HCl eluant was >95% as demonstrated by chromatography of a mixture of [3H]MTX-Gz (100 pmol, prepared enzymatically) and MTX (125 nmol) on DEAE- cellulose under standard assay conditions. Similarly, recoveries in the 0.1 M HCI eluant were shown to be >90% for the principal polyglutamylated metabolites


of AMT, folic acid, and folinic acid, as well as for the MTX polyglutamates MTX-Ga to MTX-Gs. By contrast, unmetabolized MTX and AMT were fully (>95%) eluted and folinic acid was -85% eluted with the Na acetate wash of the DEAE-cellulose columns; of the various nonpolyglutamylated substrates used in these experiments only folic acid required 0.1 M HCl for elution.

DE52 column chromatography of (anti)folate poly- ghbnates. FPGS reaction products were analyzed by gradient elution on Whatman DE52 (0.65 X 16 cm). Resin was prewashed with 0.5 M NH&l and equili- brated with 5 mM KP,, pH 7.0, using a modification of a method to resolve folic acid and its oligoglutamate derivatives (22). FPGS incubation mixtures were diluted with 0.75 ml of 93 mM Na acetate, pH 5.2, and then mixed with synthetic MTX-GI to MTX-GG standards (-20 nmol each). Samples were applied and then the columns were washed with 5 mM KPi, pH 7.0, 50 mM NaCl (70 ml) to remove unreacted L-

[3Hlglutamate. (Anti)folate polyglutamates were eluted and resolved by a linear gradient of 50 to 400 mM NaCl in 5 mM KPi, pH 7.0 (440 ml, total volume).

Fractions were analyzed for radioactivity, conductiv- ity, and &z.

HPLC analysis of MTX polyglutamatea Samples were prepared for HPLC analysis by the standard assay method except where indicated. Assay mixtures were then boiled for 15 min, filtered through a Mil- lipore Millex-HA 0.45 pm filter unit, shell frozen, and lyophilized. The residue was dissolved in the HPLC mobile phase buffer (typically 100 ~1) and a volume of 50-90 ~1 was injected [together with a mixture of unlabeled MTX-GI to MTX-G-, (-0.2 nmol each), when required] onto a Water Radial-Pak Cl8 column (0.5 X 10 cm, 10 pm particle size) developed with 0.1 M

ammonium acetate, pH 6.0, as the mobile phase [cf. Ref. (20)]. A linear gradient of 0 to 5% acetonitrile was run over 70 min at a flow rate of 1.0 ml/min, followed by an isocratic wash with 5% acetonitrile for 30 min. Fractions (0.5 ml) were collected and then analyzed for radioactivity. Unlabeled standard MTX polyglutamates were detected at 280 nm with a Waters Associates Model 440 detector.

Conjugase (y-glutamyl hydrolase) activity. Human liver FPGS was incubated with MTX-G6 (2.5 PM) or MTX-Gz (2.5 HIM) under standard assay conditions (in the absence of ATP and using unlabeled L-glutmate; these polyglutamate concentrations are comparable to those typically generated during a standard FPGS assay). Samples were prepared for HPLC analysis and tbe extent of cleavage of MTX-Ga or MTX-Gz was determined by HPLC as described above. FPGS activity was also determined under standard conditions in the presence of the conjugase inhibitor heparin (27 ng/ ml) (23), which had no effect on product formation.

Miscellaneous procedwes. Protein concentrations were determined by the Lowry method following quantitative precipitation of protein using deoxycho-

late and trichloroacetic acid (24). Kinetic parameters were derived from Lineweaver-Burk double-reciprocal plots by a weighted least-squares linear regression analysis (weighting factor = velocitf) as described previously (25). K,,, and V,,, values reported in this study are apparent kinetic constants and are ex- pressed as +SD for the best fit for six or more data points ranging from 0.2 to 5 times the observed K, values. Duplicate determinations of kinetic parameters were generally within 10% of the values quoted in the text.

RESULTS

Enzyme isolation, stability, and general properties. FPGS was purified four- to

eightfold from six different human liver samples by O-30% ammonium sulfate fractionation of a cytosolic fraction, with an activity yield of 80 f 10%. FPGS activity was assayed by the ATP-dependent incorporation of L-[3H]glutamate into a suitable (anti)folate substrate, with the radiolabeled reaction product separated from free L-[3H]glutamate by chromatography on DEAE-cellulose. Insignificant conjugase (y-glutamyl hydrolase) activity was observed in the isolated FPGS samples, with ~2% hydrolysis observed when using either MTX-Gz or MTX-GG as conjugase substrate (see Materials and Methods). Analysis of the product formed upon incubation of human liver FPGS with MTX (Fig. 1) revealed that MTX-Ga was the only radioactive product formed. Thus, catalytic assays and determinations of kinetic parameters could be performed under conditions where only a single glutamyl residue was incorporated at the r-COOH group of MTX.

Human liver FPGS was relatively stable to storage at -20°C with 74 ? 6% (n = 6 preparations) of the initial activity remaining after 5 months. Comparisons of the stability of human liver FPGS in buffer to the corresponding rat and rabbit liver enzymes under similar conditions are shown in Fig. 2.A. Human liver and rat liver FPGS both inactivated rapidly at 37°C [tl,z (HL36) = 7.5 min; tllz (HL37) = 8.5 min; t1,2 (rat FPGS) = 11 min], while the rabbit liver enzyme was considerably more stable [tr,z (rabbit FPGS) = 68 min]. Human liver FPGS was markedly stabilized by ATP [& (HL36) = 38 min at 5 mM ATP] (Fig.


800

40 80 120

Fraction Number

FIG. 1. Identification of the reaction product fromed by human liver FPGS under standard incubation conditions. HL39 FPGS was incubated with MTX for 2 h under standard assay conditions except that 20 mCi/mmol @HIglutamate was used. The complete incubation mixture was mixed with MTX-Ga to MTX-G6 standards (elution positions denoted by arrows with the designations G,) and then chromatographed on a DE52 column as described under Materials and Methods. The small radioactive peak eluting between MTX-G5 and MTX-G6 (fraction 104) was not present in the L-

[3Hlglutamate substrate and did not represent a MTX polyglutamate. This identification of MTX-Ga as the major radiolabeled product was confirmed by HPLC analysis.

9 U), with half-maximal stabilization observed at 0.18 f 0.02 mM nucleotide (data not shown). Similar stabilizations were observed for FPGS from rat and rabbit liver (data not shown). 2-ME further en- hanced the stabilization provided by ATP ItlIz (HL36) = 240 min] (Fig. 2B) with half- maximal enhancement of stability observed at 1.5 f 0.2 mM 2-ME (data not shown).

FPGS activity was negligible (~5%) in the absence of any one of the cosubstrates (L-glutamate, antifolate, or ATP) or in the absence of MgClz (presumably reflecting a high specificity of human liver FPGS for MgATP; K, (MgCl,) = 3.7 f 1.7 mM). Hu- man liver FPGS also required K+ for maximal activity (K, (KCl) = 1.6 -t 0.3 InM), with a ninefold activation obtained at 20 mM KCl. NaHC03 further activated human liver FPGS up to twofold at a concentration of 20 mM (K, (NaHC03) = 3.9 + 0.9 mM).

NaCl (20-50 mM) was without effect on FPGS activity.

Human liver FPGS exhibited a pH optimum of 8.2 to 8.4 (37°C) when 0.2 M TrisCl buffer with AMT was used as substrate {data not shown). Activity was reduced by about 50% at either pH 7.5 or pH 9, with .<lO% of the maximum activity observed

below pH 7. Negligible FPGS activity was observed at 0°C or above 40°C.

Amino acid and nucleotide cosubstrates. Kinetic analyses using L-glutamate as the variable substrate indicated an apparent K, of 1.2 rf: 0.3 mM at pH 8.3 when AMT was used as cosubstrate. Other potential amino acid substrates had no effect on the incorporation of L-[3H]glutamate into antifolate when included in the standard assay at 10 mM: L-aspartate, L-cysteine, P-glutamate, D-glutamate, L-glutamine, glutathione, and L-methionine. However, y-L-glutamyl-L-glutamine and y-L-glutamyl-L-glutamate both effected a modest but reproducible decrease in the incorporation of L-[3H]glutamate into product (10 and 20% decrease, respectively, at 10 mM dipeptide). The possibility that these apparent inhibitions might have resulted, in part, from contaminating L-glutmate was not examined.

Human liver FPGS utilized ATP as a cosubstrate, with an apparent K, of 0.25 + 0.02 mM (Fig. 3A). This value is similar to that determined for the stabilizing effect of ATP (0.18 mM, see above). Other nucleotides (5 mM) were able to replace ATP as cosubstrate with varying degrees of effectiveness. Relative specific activities of 112


= 100 t

E

(5 60 6

s g 60 .? 6 a cn 40

& 20

0 20 40 60

Preincubation The (mid

0 20 40 60 Prelncubation Tlme (mln)

FIG. 2. Time-dependent inactivation of mammalian FPGS at 3’7°C. (A) FPGS preparations from HL36 (a), HL3’7 (O), rat liver (m), or rabbit liver (0) were preincubated at 37°C for the indicated times in complete assay mixtures not containing ATP (-) or not containing antifolate (---; “+ATP”). Reactions were then initiated by the addition of 5 mM ATP or 500 pM AMT as required and residual FPGS activity was determined by the standard assay method. The 100% activity value for each FPGS preparation was determined by a standard enzyme assay without preincubation. (B) Prein- cubations and enzyme assays were performed as described in A using HL36 FPGS subject to dialysis to remove 2-ME (see Fig. 3B legend). Samples were preincubated for the times indicated in the absence of either AMT (O), AMT and ATP (m), or AMT and 2-ME (0). FPGS assays were then initiated by the addition of 500 pM AMT, 45 mM 2-ME, and/or 5 mM ATP as required. L-[3H]Glutamate was used at 20 mCi/mmol.

ATP DEPENDENCE A. - 1

2 600- .

2-W DEPENDENCE

2 B.

'a 600-

" a 200: g I' &

0 10 20 30 40 ATP (mM)

2-ME (mM)

FIG. 3. Dependence of FPGS activity on various components of the incubation mixure. (A) Incu- bations were performed under standard assay conditions using HL36 FPGS and AMT, with ATP concentrations varied as shown. (B) Incubations were performed using thiol-free HL36 FPGS (thiol removed by dialysis for 20 h at 4°C against 3 X 50 vol of 20 mM Hepes, pH 7.4, containing 0.25 M sucrose; >95% recovery of activity) diluted into assay buffer containing 2-ME to give the indicated concentrations. All other assay conditions were standard, excepting that the reaction was for 20 min and was initiated by the addition of 500 pM AMT. Data shown were corrected (0) [uncorrected data points: (o)] for the effect of 2-ME on enzyme stability. Correction factors used for this experiment were determined from an independent series of inactivation experiments (analogous to the upper two curves shown in Fig. 2B) run in the presence of 0 to 45 mM Z-ME with the correction factors based on the fraction of activity remaining after 20 min preincubation.


and 37% were observed with dATP and dGTP respectively, in place of ATP, while IJTP and dTTP showed minimal activity, and AMP was inactive. Some activity was observed using ADP, possibly due to con- tamination of our FPGS preparation by adenylate kinase [cf. Ref. (16)].

Thiol requirements. The effect of 2-ME on FPGS activity was examined after re- moval of the thiol by dialysis at 4°C (see Fig. 3B, legend). FPGS activity was then assayed in the presence of 0 to 45 mM 2- ME. An apparent 3.5-fold enhancement of activity was obtained at 45 mM 2-ME (Fig. 3B) with a 25% decrease in activity observed at 200 mM 2-ME (data not shown). A twofold stimulation was calculated from these data after correcting for the stabilizing effect of 2-ME on FPGS (see Fig. 3B, legend), with half-maximal effect observed at 2.5 f 0.4 mM 2-ME. Thus 2-ME stabilizes human liver FPGS and also enhances its catalytic activity. 2-ME (45-50 mM) was therefore included in all buffers used for enzyme isolation, storage, and assays.

Interindividual and interspecies difler- ences. The specific catalytic activities of human liver FPGS for folic acid, folinic acid, MTX, and AMT were determined using enzyme isolated from six individual liver samples (Fig. 4). The relative catalytic activities obtained for these four substrates were highly similar for each indi-

I q AMT I E3 MTX I

F L, 1200

E a

“, 600 ‘: .? 2

; 400

HL33HL35 HL36 HL37HL38 HL39 RAT RABBIT Human . Liver FPGS Liver Liver

FPGS FPGS

FIG. 4. Interindividual differences in human liver FPGS activity: comparison with rat liver FPGS and rabbit liver FPGS. Specific activities were determined using 500 pM (anti)folyl substrates under standard assay conditions with FPGS freshly prepared by 0 to 30% ammonium sulfate precipitation as described under Materials and Methods.

vidual FPGS preparation, with the order of activity being AMT > MTX > folinic acid > folic acid. Ratios of specific activities (mean + SD) for the six preparations were: AMT/folic acid = 2.18 -t 0.23, AMT/folinic acid = 1.76 + 0.29, and AMT/MTX = 1.34 + 0.16. Thus, both folates and antifolates are effectively y-glutamylated by human liver FPGS. A four- to sixfold range of FPGS activity was, however, observed among the six individuals for each of these substrates, with HL39 being the most active and HL38 the least active. Direct comparison of the human, rat, and rabbit liver enzymes at similar stages of purification indicated that although comparable activities are exhibited by each of the enzymes, a somewhat different order of substrate activity characterizes each species (Fig. 4).

Speci$city for folates and folate antagonists. Determination of the kinetic parameters exhibited by human liver FPGS with four (anti)folyl substrates evidenced a high degree of similarity between the three individual preparations studied (Table I). This suggests the presence of the same or highly similar enzyme species in each individual. Using the apparent first-order rate constant V,,,/K, for comparison (Table I, final column), the order of substrate efficiency was AMT > folinic acid > MTX > folic acid. Comparison of these data for human liver FPGS with those reported for the corresponding mouse liver enzyme [Refs. (17,26); see Table I, footnote c] revealed significant differences. Thus the relative efficiency (represented by relative V,,,/K,) of AMT as a substrate for FPGS is five to six times greater for the human as compared to the murine enzyme. Human liver FPGS also exhibited a higher V,,,,, and lower Km for MTX as compared to folic acid, while the opposite was true for mouse liver FPGS. Kinetic parameters were also determined for human liver FPGS with three reduced folates which serve as physiological substrates. Apparent Km values for dihydrofolic acid (2.8 +- 0.5 FM), tetrahydrofolic acid (7.0 + 0.4 PM), and 5-methyltetrahydrofolic acid (58 + 15 PM) (determinations using HL35 FPGS) were similar to the values reported for mouse liver FPGS (17) except that dihydrofolic acid


TABLE I

KINETIC CHARACTERIZATION OF HUMAN LIVER FPGS

Apparent FPGS Apparent V nmx Relative Relative

preparationa Substrate Kn (MM) (pmol/mg/h) v * max Vm.x/Kn”

HL33 Folic acid 137 224 210 + 31 1 1 MTX 69 + 9.2 304234 1.45 2.86 Folinic acid 5.4 ?I 0.7 227 k 26 1.08 27.4 AMT 5.1 f 0.6 576 + 50 2.74 73.1

HL35 Folic acid 101 -+16 442+54 1 1 MTX 61 f 5.2 645 + 46 1.46 2.42 Folinic acid 5.2-+ 0.6 369 + 37 0.83 16.2 AMT 4.3f 0.5 1087 f 34 2.46 57.8

HL36 Folic acid 100 2 5.1 363 -t 15 1 1 MTX 63 f 8.6 483+56 1.33 2.13 Folinic acid 3.9+ 0.5 429 f 37 1.18 30.3 AMT 3.8+ 0.4 935 f 32 2.57 68.3

Note. Apparent kinetic constants were determined for the indicated human liver FPGS preparations as described under Materials and Methods using data derived from standard assays with substrate concentrations ranging from -0.2 to 5 times the observed Km values.

a The kinetic parameters determined for human liver FPGS can be compared to the values reported for mouse liver [Refs. (17, 26)]. Apparent K,,, values: folic acid, 140 + 47 j.&M; MTX, 166 -t 49 PM; folinic acid, 8.1 + 1 pM; AMT, 17.6 + 6 pM; relative V,, /K,,, values: folic acid, MTX, 0.90; folinic acid, 16.0; AMT, 12.3.

* Calculated for each FPGS preparation by dividing the apparent V,,,, values for MTX, folinic acid, or AMT by the apparent V,, for folic acid.

’ V,.,/K,,, for MTX, folinic acid, or AMT divided by the V,,,/K, for folic acid.

exhibited a threefold lower K, value with the human liver enzyme. V,,, values relative to AMT were 0.84, 0.58, and 0.90, respectively, for the three reduced folates.

Although apparent K, values of 4 to 5 PM were exhibited by human liver FPGS with both AMT and folinic acid as substrates, a significant decrease in FPGS activity was observed as the concentration of each of these substrates was increased above 250 PM (Fig. 5 and data not shown). A similar effect was observed with the physiological reduced folates dihydrofolic acid (inhibition above 10 PM), tetrahydro- folk acid (above 50 PM), and 5-methyltetrahydrofolic acid (above 50 PM). Substrate inhibition was not observed with MTX or folic acid at concentrations ranging up to their limits of solubility (“5 mM).

Synthesis of (anti)folate polyglutamates by human liver FPGS. The distribution of oligoglutamate products of MTX was determined at a variety of substrate concen-

trations and incubation times with either human liver cytosol or partially purified FPGS. Under standard assay conditions (500 PM MTX and 1 h incubation) MTX-Gz was the only product formed by the isolated enzyme (see Fig. 1). At both 20 and 5 PM

MTX and with longer incubation times, MTX-G2 was still the major product although small amounts of MTX-G3 were formed. Higher polyglutamates were not detected (Table II). MTX-Gz was the sole detectable product formed by the cytosolic fraction, even after longer incubation times and with lower substrate concentrations, conditions presumed to be more favorable for the formation of higher polyglutamates. Similar findings were obtained with AMT as FPGS substrate, even at low drug concentrations (1 or 2 PM). In the case of folinic acid at 2.5 PM, folinic acid-G2 corresponded to the major metabolite, with ~3% of the product corresponding to folinic acid-G3. However, when the incubations


L..

MTXpolyglutawmtes as FPGS substrates. MTX undergoes extensive polyglutamylation in viva, with polyglutamates as long as MTX-GF, formed in human liver (2). The polyglutamate chain is built up stepwise (27), indicating that the individual MTX polyglutamates probably act as FPGS substrates. Steady-state kinetic analyses of MTX and four individual oligoglutamates (MTX-G2 to MTX-GS) established that the apparent V,,, values decrease significantly (>80%)as the polyglutamate chain length

-200 -100 0 100 200 is increased. Concomitantly, the apparent l/S (mM-') Km values decrease from 54 PM (MTX) to

FIG. 5. Steady-state kinetic analysis of AMT as a substrate for human liver FPGS. Incubations were

11 PM (MTX-Gr,) (Table III). Thus the low specific activity of human liver FPGS with the higher polyglutamates of MTX (data not shown) reflects a decrease in apparent V,,,,, with increasing polyglutamate chain length, a decrease that is only partially offset by a corresponding decrease in Km.

performed under standard assay conditions using HL35 FPGS and the indicated concentrations of AMT. Inhibition of FPGS activity by AMT at high concentrations is shown in the inset (data points from 125 PM to 5 mM AMT). Kinetic constants shown in this figure were determined as described under Materials and Methods using data points representing AMT concentrations from 6 to 125 pM. Essentially identical DISCUSSION

& values (& = 3.0 f 0.3 PM) were obtained in sep- arate experiments using data points representing

Mammalian liver FPGS has been iso-

AMT concentrations from 0.5 to 25 pM. lated from several species including rat, mouse, and hog liver and found to exhibit a high pH optimum, striking dependence

were carried out at 2 PM folate, up to 16% on K+, and high specificity for L-glutamate. of the product corresponded to a compound These general properties also characterize tentatively identified as folinic acid-G3. FPGS isolated from human liver cytosol,

TABLE II

FORMATION OF MTX POLYGLUTAMATES AS A FUNCTION OF INCUBATION CONDITIONS

Enzyme MTX preparation (PM)

cytoso1 20

cytoso1 20

FPGS 20

FPGS 20

FPGS 5

Incubation Total time product

(h) (pm4

2 80

20 220

2 295

20 890

20 270

Product distribution“

(%I

MTX-Ga MTX-GB

100 100

99.4 0.6

98.9 1.1 98.5 1.5

Note. Cytosol or FPGS isolated from HL39 was incubated under standard assay conditions except that L- [3Hlglutamate was used at 20 mCi/mmol and the MTX concentration and incubation time were varied as noted. Product analyses were performed by DE52 chromatography as described under Materials and Methods.

“Product distribution values are adjusted to account for the twofold higher specific activity of MTX-Ga relative to MTX-Ga. No polyglutamates higher than MTX-G3 were observed. Some 95-98% of the eluted cpm corresponded to MTX-Ga or MTX-Ga with the remainder eluted as small peaks which did not cochromatograph with any of the MTX oligoglutamate standards (see e.g., the small radioactive peak at fraction number 104 in Fig. 1).


TABLE III

KINETIC CHARACTERIZATION OF MTX POLYGLUTA- MATES AS HUMAN LIVER FPGS SUBSTRATES

Apparent Apparent KWZ V

Substrate (PM) (pmolT;g/h) (:;;::)

MTX 54 f 2.8 1025 f 51 100 MTX-G2 29 + 4.2 258 f 35 47 MTX-Gs 20 + 9.5 46k19 12 MTX-Ga 11 f 5.0 35+ 9 17 MTX-GS 11 f 2.3 39f 5 19

Note. Apparent kinetic constants were determined with HL39 FPGS as described in Table I except that L-PHIglutamate was used at 20 mCi/mmol to increase the sensitivity of the assay, and an unweighted least- squares linear regression analysis was used.

as shown in the present study. Human liver FPGS is also activated -twofold by HCO,, which has not previously been examined for its effect on FPGS activity. The ability of the human liver enzyme to utilize nucleotides other than ATP is similar to that reported for rat liver FPGS (15). Hu- man liver FPGS was found to inactivate rapidly when incubated at 37°C in the absence of ATP (tl,z - 8 min ). Although rat liver FPGS deactivated at a similar rate, rabbit liver FPGS was markedly more stable (tljB = 68 min). The stabilization of human liver FPGS by ATP observed in the present study is similar to that reported for the corresponding mouse liver enzyme (16). Finally, 2-ME was also found to im- prove FPGS stability in addition to en- hancing enzyme activity.

Human liver FPGS utilizes both folates and antifolates as substrates, with the order of reactivity (described by V,,,/K,) given by dihydrofolic acid > AMT > tetrahydrofolic acid > folinic acid > 5-methyltetrahydrofolic acid > MTX > folic acid. These differences in efficiencies are largely reflective of Km differences, which vary some 30-fold among these compounds. By contrast V,,, values varied over only a 3- fold range for the folates and antifolates studied. Although, in general, reduced folates are the preferred substrates for

FPGS, with a variety of substituents being tolerated by the enzyme (26), AMT was a better substrate for the human liver enzyme than were all the reduced folates tested except for dihydrofolic acid. This order of reactivity is distinct from that reported for mouse liver FPGS (17, 26). Di- rect comparisons of partially purified FPGS from human, rat, and rabbit liver under V,,, conditions confirmed the exis- tence of these interspecies differences. In- hibition of human liver FPGS activity by AMT and several reduced folates was observed at high (>50-250 /*M) substrate concentrations. Although inhibition of partially purified FPGS from mouse kidney and L1210 leukemia cells (but not mouse liver) has been reported in the case of folic acid (26), AMT had no inhibitory effect in those systems. Finally, although the kinetic constants determined for human liver FPGS were of the same order of magnitude as those reported for other partially purified mammalian FPGS preparations (17, 26), potentially significant differences were apparent. These findings suggest that more detailed studies of the human liver enzyme may provide a better understanding of the metabolism of antifolate drugs commonly used in cancer chemotherapy.

Human FPGS isolated from liver cytosols derived from six individuals exhibited the same relative substrate activities with four folyl derivatives. These results, together with more detailed kinetic analyses using enzyme isolated from three individuals, are consistent with the presence of the same, or at least highly similar enzyme species in each individual. A four- to sixfold interindividual range in specific activity was, however, observed. Such differences in FPGS levels, if found in viva, could contribute to interpatient differences in cytotoxicities observed during antifolate chemotherapy (28). It is possible, however, that differences in treatment of the livers prior to enzyme isolation might have con- tributed to the observed variations in FPGS levels.

Comparison of MTX with the individual polyglutamates MTX-G2 to MTX-G5 as substrates for human liver FPGS indicated


a striking decrease in substrate activity with increasing y-L-glutamate chain length. This was shown to reflect a significant decrease in V,,,, although a decrease in Km partially compensated for the reduced activity of the higher polyglutamates. A similar effect has recently been reported for a series of tetrahydrofolate polyglutamates with purified hog liver FPGS (29), although in that system the decrease in Km with increasing polyglutamate chain length is more striking, such that longer chain derivatives can be formed in vitro. By contrast, when using MTX as substrate for our human liver FPGS, polyglutamates higher than MTX-G3 were not formed. Longer polyglutamylated species such as MTX-G5 and MTX-G6 are, however, formed in vivo in human as well as in other mammalian cells (2, 4, 20). These longer polyglutamates efflux more slowly from cells than those containing fewer y-L-glutamate residues (5) and are usually present in small amounts even in cells where a lower polyglutamate is the predominant product. In the present in vitro study, however, the major product was always MTX- G:! even under conditions believed to be favorable for the formation of higher polyglutamates (low substrate concentration and long incubation time). The presence of residual conjugase (y-glutamyl hydrolase) activity was ruled out as a contributing factor by direct assay. Unfractionated human liver cytosol was also unable to cat- alyze formation of longer chain MTX polyglutamates, suggesting that the apparent inactivity of the isolated human liver FPGS preparations in chain elongation is not due to the resolution of distinct enzyme species during the purification procedure. It is, however, possible that loss or inactivation of a required activator or other regulatory factor during preparation of the cytosol may contribute to the observed inactivity. Alternatively, the greatly reduced activity of FPGS with the longer MTX polyglutamates may indicate that the in vifro reaction conditions employed in this study to maximize diglutamate formation are unsuitable for the formation of higher polyglutamates. Studies are currently un-

derway to address these and related questions in greater detail.

ACKNOWLEDGMENTS

We thank Dr. A. Rosowsky for many useful discus-

sions, Dr. J. Wright for assistance with analyses of

reduced folates, D. Trites for assistance with the

HPLC analyses, Dr. F. P. Guengerich for provision of

liver cytosols, and D. Clare and B. Chen for their care-

ful typing of the manuscript.

REFERENCES

1. MCGUIRE, J. J., AND BERTINO, J. R. (1981) Mo1 Cell

Biochem. 38,19-48. 2. KRAKOWER, G. R., NYLEN, P. A., AND KAMEN,

B. A. (1982) Anal. Biochem. 122,412-416. 3. SHANE, B. (1982) Amer. J Clin. Nutr. 35,599-608. 4. PRIEST, D. G., HAPPEL, K. K., MANGUM, M., BED-

NAREK, J. M., DOIG, M. T., AND BAUGH, C. M.

(1981) Anal. Biochem. 115,163-169.

5. MCBURNEY, M. W., AND WHITMORE, G. F. (1974)

Cell 2, 173-182.

6. BALINSKA, M., GALIVAN, J., AND COWARD, J. K.

(1981) Cancer Res. 41,2751-2756.

7. WHITEHEAD, V. M. (19’77) Cancer Res. 37.408-412. 8. MATHERLY, L. H., FRY, D. W., AND GOLDMAN,

I. D. (1983) Cancer Rex 43,2694-2699. 9. ALLEGRA, C. J., CHABNER, B. A., DRAKE, J. C., LUTZ,

R., RODBARD, D., AND JOLIVET, J. (1985) J. BioL Chem. 260,9720-9726.

10. ALLEGRA, C. J., DRAKE, J. C., JOLIVET, J., AND

CHABNER, B. A. (1985) Proc. N&L Acad. Sci. USA 82,4881-4885.

11. SHANE, B. (1980) J. BioL Chem. 255,5655-5662. 12. BOGNAR, A. L., AND SHANE, B. (1983) J. BioL Ch.em

258,12574-12581.

13. BOGNAR, A. L., OSBORNE, C., SHANE, B., SINGER,

S. C., AND FERONE, R. (1985) J. BioL Chem. 260, 5625-5630.

14. CHICOWICZ, D., COOK, J., GEORGE, S., AND SHANE,

B. (1985) in Proceedings of the Second Work-

shop on Folyl and Antifolyl Polyglutamates

(Goldman, I. D., Ed.), pp. 7-13, Praeger, New

York.

15. TAYLOR, R. T., AND HANNA, M. L. (1977) Arch. Biochem. Biophys. 181,331-344.

16. MCGUIRE, J. J., HSIEH, P., COWARD, J. K., AND

BERTINO, J. R. (1980) J. BioL Chem. 255, 5776- 5788.

17. MORAN, R. G., AND COLMAN, P. D. (1984) Biochem- istry 23,4580-4589.

18. PRISTUPA, 2. B., VICKERS, P. J., SEPHTON, G. B.,

AND SCRIMGEOUR, K. G. (1984) Canad J.

B&hem. Cell Biol. 62,495-506. 19. FREI, E., III, ROSOWSKY, A., WRIGHT, J. E., CUCCHI,

C. A., LIPPKE, J. A., ERVIN, T. J., JOLIVET, J.,


AND HASELTINE, W. A. (1934) Proc. Natl. Acad. Sci USA 81,2873-2877.

20. ROSOWSKY, A., WRIGHT, J. E., CUCCHI, C. A.,

LIPPKE, J. A., TANTRAVAHI, R., ERVIN, T. J., AND

FREI, E., III (1985) Cancer Res. 45,6205-6212. 21. NYLEN, P. A., ABELSON, H. A., WHITEHEAD, M. V.,

DOLNICK, B., PETERSON, D. W., AND KAMEN,

B. A. (1984) Proc. Amer. Assoc. Cancer Res. 25, 309.

22. BAUGH, C. M., AND KRUMDIECK, C. L. (1971) Ann. N. I: Acad. Sci. 186,7-28.

23. SILINK, M., REDDEL, R., BETHEL, M., AND ROWE,

P. B. (1975) J. Biol. C&m. 250,5982-5994.

24. BENSADOUN, A., AND WEINSTEIN, D. (1976) Anal. B&hem. 70,241-250.

25. WAXMAN, D. J., AND WALSH, C. (1983) Biochemistry 22,4846-4855.

26. MORAN, R. G., COLMAN, P. D., ROSOWSKY, A.,

FORSCH, R. A., AND CHAN, K. K. (1985) MoZ. Phurmucol 27,156-166.

27. MORAN, R. G., COLMAN, P. D., FORSCH, R. A., AND

ROSOWSKY, A. (1984) J. Mea! Chem 27, 1263-

1267.

28. JOLIVET, J., COWAN, K. H., CURT, G. A., CLENDEN-

INN, N. J., AND CHABNER, B. A. (1983) N. EngZ. J. Med 309,1094-1104.

29. SHANE, B. (1986) in Chemistry and Biology of

Pteridines (Cooper, B. A., and Whitehead,

V. M., Eds.), pp. 719-728, de Gruyter, Berlin.

Documents

Human liver folylpolyglutamate synthetase: Biochemical characterization and interactions with folates and folate antagonists